a) Genetic labeling of cranial nerves and other projection sensory neurons (Sajgo et al 2016). Our description of Brn3 combinatorial codes in the Dorsal Root Ganglia and inner ear raised the possibility that Brn3s provide a general combinatorial code for projection neurons of all major sensory pathways. We completed our understanding of this code by performing a developmental profile of Brn3b expression in cranial nerves and nuclei of the brainstem. We find that Brn3b is dynamically expressed in the somatosensory component of cranial nerves II, V, VII, and VIII and visceromotor nuclei of nerves VII, IX, and X as well as other brainstem nuclei, but that no olfactory or taste pathways are positive for Brn3s. Thus, in the mouse Brn3 combinatorial codes are utilized by all major sensory pathways with the exception of taste and smell. These findings further strengthen the link between Brn3 transcription factors and the major sensory pathways, suggesting a marking of the ascending sensory pathway by Brn3s. Together with previous work, it appears that at the level of cranial nerves, Brn3s are participating in sensory and visceral pathways, Phox2 transcription factors mark mostly visceral pathways, while Islet transcription factors define motor neurons and sensory projection neurons. These findings are interesting from an evolutionary perspective, but have also practical implications, as molecular mechanisms as well as tools and strategies can be shared between the various systems to refine our circuit analysis in the visual system. As an example of such cross-talk, in our collaboration with Drs. Niu and Luo at U.Penn, we used our genetic approaches to define the distribution of cRet+ mechano - and Parvalbumin+ proprioceptive projection sensory neurons in the dorsal columns of the spinal chord (Niu et al 2013). The sensitivity of our AP reporters allowed our collaborators to follow the centrally projecting axons of these somatosensory neurons throughout the entire spinal chord to their targets in the brain stem. b) Gene Expression profiling of Brn3AP RGCs and their brain targets by RNASeq (Sajgo et al, Manuscript under submission) In order to determine the molecules expressed in RGCs at critical stages for their axon, dendrite and synapse formation (E15 and P3), we have established an immunoaffinity purification strategy based on our Brn3CKOAP alleles. This was done by magnetic separation of Brn3AP RGCs, using monoclonal antibodies against AP, a GPI linked surface molecule, followed by RNA extraction and RNA sequencing in the N-NRL genomics core. Using this strategy, we identified (i) genes enriched in RGCs compared to the retina, (ii) genes enriched in Brn3aAP vs Brn3bAP RGCs (presumed to convey the specific features distinguishing RGC cell types), and (iii) genes up- or downregulated by loss of Brn3a or Brn3b (potentially mediating the developmental programs that the Brn3s are thought to directly or indirectly regulate in RGCs). About 3000 transcripts are enriched in RGCs when compared to the retina, and about 900 transcripts are found to be regulated by Brn3b and/or Brn3a. We find that Brn3b loss affects RGCs significantly more than Brn3a loss. Moreover, the transcripts selectively expressed in RGCs, or affected by Brn3b loss differ significantly between E15 and P3 RGCs, suggesting distinct expression profiles for the different stages of development. We validated our RNA sequencing screen, by in situ hybridization on P3 eye sections, and find that more than half of the 233 targets are actually RGC specific. A similar ratio was found for 265 genes that were confirmed by queries of the Allen Institute In Situ Hybridization atlas of E15 mouse embryos. We complemented these RGC profiles with RNAseq analysis of retinorecipient brain nuclei at P3, an age when RGC axons have reached the target and synapse formation is extremely active, and report our findings in the same study. Amongst the many differentially expressed genes in our screen, we have focused on two classes of molecules: (i) transcription factors, (ii) adhesion molecules/signaling receptors. We find that, whereas about 60 % of all annotated transcription associated genes are expressed in RGCs, 322 genes are enriched in RGCs compared to retinas, and 95 are under Brn3 control. These transcription factors, some of which had already been described, generate a wide combinatorial code demonstrating specificity for distinct RGC subpopulations. Interestingly a much smaller number of adhesion molecules and guidance receptors are enriched in our RGC populations (about 200 transcripts). To address the potential functions of molecules identified in our screen, we overexpressed a small subset (10 genes) in HEK293 cells, and find that several of them have the capacity to induce membrane processes reminiscent of neurites. It could therefore be that cell-autonomous mechanisms participate in determining neuronal arbor formation, that is then further refined and positioned by transmembrane receptors/adhesion molecules mediating cell-cell and/or cell-matrix interactions. This novel way of thinking about neuronal arbor formation will drive our research moving forward. Using a Cre dependent, AAV-based overexpression approach to determine the subcellular localization of some of our targets in vivo in Brn3bCre RGCs, we find distributions consistent with roles in vesicle trafficking within neurites or at synapses. Finally, out of a set of 79 genes proposed to be associated with Glaucoma (Rev1, Rev2) in human genetics studies, only 12 appeared to be enriched RGCs, potentially revealing molecular pathways associated with susceptibility to RGC damage. Thus, our RGC gene expression profiling project promises to be an extremely useful resource for the visual system field and in general for the neuroscience community. c) Dre and roxP tools for intersectional genetic manipulations (Chuang et al 2016) Previously generated Brn3aCKOCre; ROSA26AP and Brn3bCKOCre; ROSA26AP control retinas and brains showed a small but significant background recombination effect, most likely due to incomplete transcriptional termination at the STOP signal. Thus, although the proof of principle concept was successful, the conditional lines are not optimal for the desired application. We therefore are developing analogous mouse lines based on inversion-excision strategies, which will bypass the read-through problem. These strategies require novel rox (Dre target) sites that are incompatible with the wild type roxP, but efficiently recombine with each other. Since none were available, we performed a screen for such sites using a synthetic degenerate rox site library and a bacterial selection system, and validated the sites in E. Coli (prokaryotic), HEK293 (eukaryotic) systems, and by biochemistry with purified components. The novel rox sites we identified show essentially no recombination with the wild type, about 50 % recombination efficiencies with themselves when compared to wild type roxP, and no cross-reactivity with Cre in either E. coli or HEK-293 cells. We then created an expression vector carrying a Dre-rox inversion-excision cassette analogous to the Cre-lox cassette used in the FLEX approach and successfully tested it in HEK393 cells. d) A novel visual behavior apparatus allows the direct comparison of eye (Optokinetic) and head compensatory movements in mice (Kretschmer et al 2015, Wang et al 2016). In order to evaluate the functional consequences of our genetic manipulations of RGC subpopulations or cell types, we have built a multi-purpose visual behavior apparatus capable of measuring Pupillary Light Reflex, Eye movements in restrained or freely behaving mice (Optokinetic Response), and head movements in freely behaving animals (Optomotor Response).